JHR 90: 129-152 (2022) age JOURNAL OF = 4 Pe tevewed open-scoass Journal doi: 10.3897/jhr.90.75807 RESEARCH ARTICLE () Hymenopter a g https://jhr.pensoft.net Thelmternaonl Sciey of Hymenopexriss, RESEARCH Integrative taxonomy based on morphometric and molecular data supports recognition of the three cryptic species within the Encyrtus sasakii complex (Hymenoptera, Encyrtidae) Andrey Rudoy', Chao-Dong Zhu'*?, Rafael R. Ferrari', Yan-Zhou Zhang' | Key Laboratory of Zoological Systematics and Evolution, Institute of Zoology, Chinese Academy of Sciences, 1 Beichen West Road, Chaoyang District, Beijing 100101, China 2 University of Chinese Academy of Sci- ences, 19A Yuquan Road, Shijingshan District, Beijing 10049, China 3 State Key Laboratory of Integrated Pest Management, Institute of Zoology, Chinese Academy of Sciences, 1 Beichen West Road, Chaoyang District, Beijing 100101, China Corresponding author: Yan-Zhou Zhang (zhangyz@ioz.ac.cn) Academic editor: Petr Jansta | Received 28 September 2021 | Accepted 16 March 2022 | Published 29 April 2022 Attp://zoobank. org/S2BFC574-91 CC-472E-AB4B-0095 DEE349A0 Citation: Rudoy A, Zhu C-D, Ferrari RR, Zhang Y-Z (2022) Integrative taxonomy based on morphometric and molecular data supports recognition of the three cryptic species within the Encyrtus sasakii complex (Hymenoptera, Encyrtidae). Journal of Hymenoptera Research 90: 129-152. https://doi.org/10.3897/jhr.90.75807 Abstract Morphometrics has established itself as one of the most powerful tools for species delimitation, particularly for morphologically-conserved groups of insects. An interesting example is the parasitoid Encyrtus sasakii Ishii (Hymenoptera: Chalcidoidea: Encyrtidae), which was recently subdivided into three cryptic species that are seemingly well-delimited with the available DNA data but nearly indistinguishable morphologi- cally. Here, we performed linear morphometric analyses of the antenna as well as shape analyses of the ovi- positor and hypopygium (the last two are key structures associated with host location and selection) to shed further light on the taxonomic status of the E. sasakii complex. Principal component analyses were carried out to visualize the amount and direction of shape variation in the ovipositor and hypopygium. Comple- mentarily, we constructed phylogenetic trees using a Bayesian approach based on two markers (28S and COI). We found statistically-significant differences in the relative size of the funicle and of the two proximal claval antennomeres among the three species. Our analyses also indicated that the outer plates of the ovipos- itor show remarkable allometric changes and that both the stylus and shield of the ovipositor are relatively well conserved among species. We nonetheless found consistent interspecific differences in the shape of the 2™ outer plate of the ovipositor and hypopygium. Also, both our COI and combined trees recovered three strongly-supported major clades, each corresponding to one of the three cryptic species. We discuss that changes in the shape of the ovipositor may have played an important role in host shift and speciation within Copyright Andrey Rudoy et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. 130 Andrey Rudoy et al. / Journal of Hymenoptera Research 90: 129-152 (2022) the E. sasakii complex. Even though the recent descriptions of both E. eulecaniumiae Wang & Zhang, 2016 and E. rhodococcusiae Wang & Zhang, 2016 appear not to fully satisfy the International Code of Zoological Nomenclature, a simple resolution for the sake of taxonomic stability is proposed herein. Keywords Allometry, Chalcidoidea, DNA barcoding, morphometrics, parasitoid, species delimitation Introduction Cryptic species are phylogenetically closely-related ones that exhibit no unambiguous morphological differences to readily permit their distinction (Tan et al. 2010; Pramual et al. 2011; Lukhtanov 2019). This understanding has led researchers to use additional lines of evidence — such as behavior (Voda et al. 2015), physiology (Allemand 1982) and ecology (Amato et al. 2007) — as an attempt to more reliably identify cryptic species (Dépraz et al. 2009; Orr et al. 2020). Obtaining such kind of data, however, requires either observation in field or experimentation with laboratory-reared animals, which is usually impeditive for most groups of organisms (Milne et al. 2007). Thus, the use of tree inference methods in combination with DNA data (which can today be obtained from fairly-old museum specimens) has arguably become the most popular approach for cryptic species identification (e.g., Padula et al. 2016; Delicado et al. 2019), especially after the onset of the ‘DNA barcoding’ era (Hebert et al. 2003a, b). In spite of important advances in the field of molecular taxonomy (see Gokhman 2018), comparative morphology indisputably remains as a reliable approach to species iden- tification and delimitation within Hymenoptera (e.g., Zhang et al. 2014; Monckton 2016; Ferrari 2017; Nugnes et al. 2017; Hasson and Schmidt 2020). Morphological taxonomists have been adopting multiple different approaches as an attempt to better solve the puzzling problem of cryptic species. A notable one has been inferring gaps in the morphological variation exhibited by closely-related species across their geographical ranges (Zapata and Jiménez, 2011). However, it has been demonstrated that continuous variation may be interrupted by factors that are not necessarily associated with diversification (Viggiani 1999), meaning that discontinu- ity is not always an evidence for speciation (Gutiérrez-Valencia et al. 2017). Another popular approach for cryptic-species delimitation is geometric morphometrics, which is based on statistical analyses of landmark data (Sproul and Maddison 2017). Mor- phometric studies have largely relied on two premises: first, allometric changes arise from directional and non-symmetrical selection of morphological traits (Pomfret and Knell 2006), including non-sexual ones (Frankino et al. 2007); second, such selection ultimately leads to speciation (Soto et al. 2007). Intraspecific allometry may be as- sociated with either environmental (Klingenberg and Zimmermann 1992) or genetic factors (Trotta et al. 2010), although a combination of the two appears more likely. Both the antenna and the ovipositor apparatus are structures of singular taxonomic relevance for parasitoid wasps. Females utilize the former as a chemical radar to search Integrative taxonomy of the Encyrtus sasakii complex 131 for hosts in the environment (Broad and Quicke 2000). This structure is also used by both sexes for intraspecific recognition during copulation (Arbuthnott and Crespi 2018). This makes the antenna vital not only for the reproduction of parasitoid wasps, but ultimately for their environmental adaptation (Fea et al. 2019). As a result, selective pressures have produced a myriad of different antennal forms (Krishnan and Sane 2015), which can be used as frameworks to understand the diversity of parasitoid wasps across different hierarchical scales. Female parasitoids use the ovipositor to locate and lay eggs inside their hosts (Vinson 1998). It is known that host specialization, including changes in the egg-laying behavior, plays an important role in the evolution of the ovipositor in parasitoids (Ghara et al. 2011; Desneux et al. 2012). Morphological changes in this key structure, in turn, may result in prezygotic barriers for conspecific populations (Rull et al. 2013). In fact, host specialization has been pointed out as a major driver of speciation in parasitoid wasps (Zhang et al. 2011; Chesters et al. 2012; Qin et al. 2018). The Encyrtus sasakii Ishii complex comprises three parasitoid species (E. sasakii, E. eulecaniumiae Wang & Zhang and E. rhodococcusiae Wang & Zhang) of coccoid scale insects (Wang et al. 2016). While E. sasakii was described almost a century ago (Ishii 1928), both E. eulecaniumiae and E. rhodococcusiae were only recently identified (Chesters et al. 2012) and then formally described (Wang et al. 2016). All three species are seemingly endemic to China and Japan: EF. eulecaniumiae and E. rhodococcusiae occur sympatricly while E. sasakit is allopratic in relation to the other two (see Wang et al. 2016: Suppl. mate- rial 1: Fig. S3), although no impassible geographic barriers separating them seem to exist. Species of the E. sasakii complex are almost indistinguishable when only traditional morphological methods are used (Chesters et al. 2012). They are nonetheless signif- cantly divergent according to the available molecular data, which showed that E. sasakii and FE. rhodococcusiae are more closely related with each other than either of them is in relation to E. eulecaniumiae (Wang et al. 2016). Laboratory-reared individuals did not exhibit any interspecific courtship and mating behavior during controlled experiments, which further supported the view that they should be treated as different species. How- ever, linear morphometric analyses carried out by the same authors did not detect any unambiguous morphological differences among the three species to support their mo- lecular data (Wang et al. 2016). Thus, the main objective of this paper is to shed further light on the taxonomic status of the encyrtid species of the E. sasakii complex through an integrative approach to taxonomy. Specifically, we present the results of phyloge- netic analyses of two molecular markers and morphometric analyses of the antenna, ovipositor and hypopygium of females and discuss their taxonomic implications. Methods Specimens examined The material examined includes recently collected specimens plus part of those studied previously by Wang et al. (2016). All specimens (including all types) are deposited in £32 Andrey Rudoy et al. / Journal of Hymenoptera Research 90: 129-152 (2022) the entomological collection of the Institute of Zoology, Chinese Academy of Sciences, Beijing, China (IZCAS). Additional collection data are given in Suppl. material 1: Table S1. Analysed structures and imaging This study is based on morphometric analyses of the antenna, ovipositor and hypopyg- ium of female parasitoids (Fig. LA—C). Funicular and claval antennomeres are herein abbreviated as F and C, respectively, followed by appropriate number (for example, F1 for the first funicular antennomere). All analysed structures were dissected from previ- ously-relaxed specimens using fine-tipped forceps and then mounted onto microscopy slides in Canada balsam. Digital images of slide-mounted structures were obtained with a Leica DM 2500 microscope equipped with a Canon EOS 550D camera, using a magnification of 100x (except the forewing that was imaged with a magnification of Figure |. Indications as to how structures were measured (red lines) and fixed landmarks defined (red stars) A ovipositor of E. eulecaniumiae (1 — stylus, 2 — ovipositor shield, 3 — 1st outer plate, 4 — 2nd outer plate) B hypopygium of E. rhodococcusiae C antenna of E. rhodococcusiae D wing of E. eulecaniumiae. Scale bar: 400 pm. Integrative taxonomy of the Encyrtus sasakii complex 133 50x). For each structure, 10 to 20 photographs were taken and then stacked together to produce multifocus images using the program Helicon Focus v.3.10.3. In Encyrtus, the funicle comprises six well-delimited antennomeres (F1—F6), while the clava consists of three (C1—C3) nearly-fused ones (Fig. 1C). The ovipositor typi- cally has the same general shape (Figs 1A, 2A—D), exhibiting no clear structural differ- ences within the genus. The stylus (Fig. 2A) forms the central axis of the ovipositor and is composed of the 1* and 2™ valvulae that are firmly connected to one another. The 3" valvula (gonostylus) and 2°¢ valvifer are completely fused in Encyrtus and together they form the ovipositor shield (Fig. 2B), which surrounds the stylus laterally. The 1* valvi- fer, which is connected to the 2" valvifer and both outer plates basally (Noyes 2010), was not included in our morphometric analyses due to the absence of well-delimited margins. Here, the 2" outer plate refers to the ventrally-oriented, subtriangular struc- ture that is somewhat loosely connected to the external margin of the 1“ outer plate (Fig. 2C, D). The hypopygium consists of the seventh metasomal sternite (Fig. 1B). Measurements and linear morphometrics Image files were imported in Image] v. 1.8 (National Institutes of Health; available at http://imagej.nih.gov/ij/), where the measurements used in our linear morphometric D Figure 2. Indication as to how semi-landmarks (red stars) were marked on the various components of the ovipositor A stylus (1 and 2™ valvulae) B shield (1 and 2" valvifers) C 1“ outer plate D 2"! outer plate. Blue and red lines depict reference baselines and distances to them from the landmarks, respectively; black rectangles represent the structures prior to size transformation. Scale bar: 400 micrometers. 134 Andrey Rudoy et al. / Journal of Hymenoptera Research 90: 129-152 (2022) analyses were obtained. To minimize errors, all measurements were repeated twice and the resulting averages were then used as input values. In this paper, forewing length was used as a proxy for body size based on prelimi- nary analyses and previous entomological studies (e.g., DeVries et al. 2020; Wongler- sak et al. 2020). The forewing was measured as the distance between the basal point of the wing membrane and the tip of the wing; its maximum width was measured through the perpendicular line connecting the apexes of the postmarginal vein and posterior margin of the forewing (Fig. 1D). The total length of the antennal clava was obtained by summing up the maximum lengths of each of its three antennomeres; its maximum width was measured along the sulci that separate the claval antennomeres (Fig. 1C). Following Wang et al. (2016), we calculated the total length of the funicle as the sum of the length of each funicular antennomere, using the sensilla as a refer- ence point; the maximum width of the funicle was taken as the width of its distalmost antennomere (i.e., F6). The length of the pedicel was taken as the maximum distance between its articulation with the scape and its upper margin. The scape was measured as the distance between its lower margin (near the antennal socket, i.e., including the radicle) and its articulation with the pedicel. All measurements were obtained from the dorsal surfaces of the antennal parts, except the clava, which was measured from both the dorsal and ventral surfaces. The total length of the antenna was then calculated as the sum of the lengths of the clava (dorsal surface only), funicle, pedicel and scape (Fig. 1C). The total length of each analysed component of the ovipositor was meas- ured as the distance between their anterior- and posteriormost sclerotized parts; their maximum width was taken through the perpendicular line connecting their inner- and outermost sclerotized parts (Fig. 2A—D). As the antennomeres of the three studied species of Encyrtus have well-conserved shapes, they were not included in our shape analyses (vide infra). Rather, we carried out statistical analysis of variance (ANOVA) using the Remdr package (Fox 2005) in the program R (R Core Team 2019) to see whether there was significant difference in the length, width and corresponding length/width ratio (henceforth LW ratio) of each antennal antennomere. Additionally, we calculated the slope of regression of the LW ratio of each antennal antennomere. The aforementioned test was also carried out with each part of the ovipositor separately. Shape analyses We performed shape analyses of the hypopygium (Fig. 1B) and ovipositor (Fig. 2A—D). Prior to landmarks digitalization, we generated separate image files for the stylus, shield, and 1 and 2™ outer plates of the ovipositor using the previously-obtained pictures of the entire ovipositor. The images were then converted to the same size (500, 750 and 1,000 pixels for both the stylus and 2™ plate of the ovipositor, 1" plate of the oviposi- tor and ovipositor shield, respectively) while maintaining their original proportions (Fig. 2A—D) in the program PAST v.3.2 (Hammer et al. 2001). To ensure homology among the landmarks set subsequently, we fixed the images of the ovipositor into the Integrative taxonomy of the Encyrtus sasakii complex 135 same position using the following well-defined reference points: central axis of stylus; connection between the 1* valvifer and the lateralmost point of the ovipositor shield; two points of the apical edge of the outer plate; connection between the 1“ valvifer and the lateralmost point of the 24 outer plate. The images of the hypopygium were fixed based on its well-defined lateral concavities. We used Image] to digitalize seven fixed landmarks on the hypopygium (Fig. 1B) and 54 semi-landmarks on the ovipositor (Fig. 2A—D); the latters were distributed as follows: stylus (n = 11, Fig. 2A), ovipositor shield (n = 20, Fig. 2B), 1* outer plate (n = 10, Fig. 2C) and 2" outer plate (n = 13, Fig. 2D). The number of semi-landmarks varied according to the shape complexity of each ovipositorial component. More spe- cifically, we marked equidistantly distributed points along the reference levels, cover- ing the outlines of all structures entirely, except for the often crumpled apex of the 24 outer plate. However, the soft tissue surrounding more strongly-sclerotized parts of both ovipositor and hypopygium were not considered. The selected landmarks were superimposed using the Procrustes method (Rohlf 1990) in PAST to minimize the influence of size and position on subsequent analysis. We then performed evolutionary and ordinary principal component (EPC and PC, respectively) analyses to visualize the amount and direction of shape variation in the ovipositor. EPC analysis was carried out to control for the influence of phylogenetic inertia on our morphometric data, using a newly-constructed Bayesian tree as input (vide infra). Correlation analyses We performed ordinary correlation analyses in PAST and used the PDAP: PDTREE package v.1.15 (Midford et al. 2010) available in Mesquite 3 (Maddison and Mad- dison 2019) for phylogenetically constrained correlation analyses. Logarithm-trans- formed values of the following measurements were used in the calculation of allometry coeficients: (i) width and total length of the antenna; (ii) length and width of each funicular antennomere separately; (iii) length and width of all studied components of the ovipositor; (iv) and forewing length. We tested for correlation among the shape parameters of each component of the ovipositor and hypopygium using evolutionary and ordinary principal components. We also carried out ANOVA analyses to test for the significance of the slopes of regression among species. DNA dataset and phylogenetic inference The DNA dataset employed in this study consisted of partial sequences of two loci: 28S rDNA large subunit (28S) and cytochrome c oxidase subunit I (COD). Most of the sequences of both 28S (56 of 87, c. 64%) and COI (59 of 87, c. 68%) were origi- nally generated by Chesters et al. (2012) and obtained from GenBank; the remaining sequences were newly generated (see Suppl. material 1: Table S1). Field collected para- sitoids were euthanized with 95% ethanol and then stored at -20 °C freezer until fur- ther processing. DNA extraction, amplification and sequencing were then performed 136 Andrey Rudoy et al. / Journal of Hymenoptera Research 90: 129-152 (2022) through the protocols outlined in Chester et al. (2012). To cancel the well-known problem associated with missing data (see Wiens 2003), we only added to our phylo- genetic analysis the terminals for which both 28S and COI sequences were available. The obtained sequences of 28S and COI were pooled within two separate sequence blocks, which were then aligned separately in BioEdit v.7.04.1 (Hall 1999) using the ClustalW algorithm (Thompson et al. 1994). The resulting alignments were visually inspected and any obvious misalignments manually corrected. The longest sequences of each block were trimmed so that all regions were present in most samples. The two individual blocks (28S = 566 bp, COI = 601 bp) were concatenated in Mesquite result- ing in a single multilocus matrix of 1167 bp in length. In total, we conducted three Bayesian phylogenetic analyses with our molecular dataset: two separate single-locus analyses (28S and COI) and one multilocus analysis. In all cases, 87 terminals were included, 75 of which belonging to the E. sasakii com- plex (ingroup). The remaining 12 terminals were outgroups belonging to three differ- ent species, namely: E. aurantii (Geoffroy, 1785) (four terminals), EF. infelix (Embleton, 1902) (five terminals) and E. infidus (Rossi, 1790) (three terminals). The analyses were performed remotely in CIPRES Science Gateway (Miller et al. 2011) using MrBayes v.3.2.6 (Huelsenbeck and Ronquist 2001; Ronquist et al. 2012). First, we used Par- titionFinder v.2.1.1 (Lanfear et al. 2017) to choose the best partitioning scheme and model(s) of nucleotide substitution for our data, using the Bayesian Information Cri- terion and “greedy” option. The following partition and models were determined by PartitionFinder and then implemented in the phylogenetic analyses: 28S — single parti- tion (SYM +1+T); COI-— 1* and 2% codons (GTR +1+T), 3% codon (GTR+14T). In MrBayes, each analysis consisted of two four-chained MCMC runs for 100 million generations, with sampling of model parameters occurring every 1,000 generations. The initial 10% of generations were discarded as burin. The program Tracer v.1.7.1 (Rambaut et al. 2018) was used to verify whether the two independent MCMC runs of each analysis had converged and whether the model parameters had reached and remained at the stationary phase. Majority-rule consensus trees were then annotated in Fig Tree v.1.4.4 (Rambaut 2016). Results Taxonomy Encyrtus eulecaniumiae Wang & Zhang Encyrtus eulecaniumiae Wang & Zhang in Wang et al. (2016) (nomen nudum) Diagnosis. Encyrtus eulecaniumiae can be morphologically differentiated from its clos- est allies by having the 2¢ outer plate at least 0.65x as long as the ovipositor shield (2°¢ outer plate less than 0.60x as long as the ovipositor shield in both E. sasakii and Integrative taxonomy of the Encyrtus sasakii complex 137 E. rhodococcusiae). Encyrtus eulecaniumiae can be further distinguished from E. rhodoc- occusiae by its shallowly concave hypopygium (hypopygium deeply concave in E. rho- dococcusiae). According to Wang et al. (2016), E. eulecaniumiae is molecularly distinct from both £. sasakii and E. rhodococcusiae in having the following nucleotides in the COI marker (according to Wang et al. 2016): 14 (A), 47 (G), 53 (G), 68 (G), 134 (A), 203 (G), 236 (G), 266 (G), 271 (G), 281 (A), 341 (A), 461 (A), 470 (A) and 533 (A). Encyrtus eulecaniumiae is also unique ecologically by being the only one to attack the coccids Eulecanium kuwanai Kanda and E. giganteum Shinji. Description. (reproduced from Wang et al. 2016) Female — Length about 3 mm including ovipositor sheaths. Colouration: Head black around ocellar area, anterior ocellus to top of scrobe pale brownish yellow to black, malar space brown; face largely brownish yellow; radicle pale orange to brown; scape mostly pale brown (Wang et al. 2016: fig. 6a); pedicel dark brown to black; funicular segments black; clava black; pronotum dorsally dark brown, laterally pale brown; thorax covered with dark brown setae, mesoscutum mostly black dorsally, laterally brown; scutellum black with a broad transverse yellow band and tuft of black bristles apically; metanotum black; tegula brown; mesopleuron pale brown; fore and hind coxae brownish yellow; mid coxa mostly brown; legs otherwise brown; basal one third of forewing hyaline, infuscate elsewhere; forewing with a series of long bristles just below the apical third of submar- ginal; propodeum brown dorsally, yellowish brown laterally; gaster black; ovipositor sheaths yellow, except apical one-third dark brown (Wang et al. 2016: fig. 6b). Head: Frontovertex about half the head width; ocelli forming an obtuse angle (about ~ 120°); scrobal depression /-shaped in frontal view; eye at least superficially bare; torulus separated from mouth margin by about one of its own length; toruli separated from each other by about 4x their own diameter. Antenna (Wang et al. 2016: fig. 6a) 13-seg- mented, scape cylindrical, 4.5x as long as broad; F1 about 1.4x longer than wide, F2 a little longer than wide, F3 and F4 as long as wide, F5 and F6 a little wider than long; clava short and 3-segmented; clypeus with three to six long, suberect setae; maxillary and labial palpi with 4 and 3 segments respectively. 7horax: Pronotum about one- sixth of mesoscutum length in dorsal view, with surface sloping from posterior margin, uniformly setose and fine reticulate sculpture. Mesoscutum 1.47x wider than long; uniformly convex, setose and finely reticulated, without notauli. Scutellum with a tuft of black bristles apically. Side of propodeum nearly bare below spiracle, but with a few inconspicuous setae on posterior margin above hind coxa; mesotibia with strong setae apically, without differentiated rows of spines; mesotibial spur about 1.7x as long as apical width of tibia (Wang et al. 2016: fig. 6e); mid basitarsus about 2x as long as wide and nearly as long as remaining segments. Forewing (Wang et al. 2016: fig. 6c) hya- line proximally, distinctly infuscate on apical two thirds; stigmal vein apically curved. Gaster: broadly sessile, with seven visible, uniformly gastral tergites; hypopygium al- most reaching apex of gaster. Type material. (reproduced from Wang et al. 2016): Holotype — 2, specimen E4- 306A, China, Beijing, Shijingshan (Badachu Park), 15.V.2014, col. Ying Wang, ex. Eulecanium kuwanai on Ulmus pumila (deposited in IZCAS). Paratypes — 22.2, Chi- 138 Andrey Rudoy et al. / Journal of Hymenoptera Research 90: 129-152 (2022) na, Beijing, Shijingshan (Badachu Park), 15.V.2014, col. Ying Wang, ex. Eulecanium kuwanai on Ulmus pumila (deposited in IZCAS); 42 9 243, China, Heilongjiang, Harbin, 9.V1.2011, col. Ying Wang, ex. Eulecanium kuwanai on Ulmus pumila (deposited in IZCAS); 1629 53, China, Henan, Zhengzhou, 4,.V.2007, col. Xiong Wang, ex. Eulecanium kuwanai on Sophora japonica (deposited in IZCAS); 2219 2 7246, China, Inner Mongolia, Hohhot, 26.V.2012, col. Haibin Li, ex. Encyrtus giganteum on Sophora japonica (deposited in IZCAS); 622.9 4033, Shanxi, Taiyuan, 1.V.2007, col. Jie Li, ex. Eulecanium kuwanai on Siophora japonica (deposited in IZCAS); 8599 3335, China, Shandong, Taian, 11.V.2008, col. Yan- zhou Zhang, ex. Eulecanium kuwanai on Sophora japonica (deposited in IZCAS). Encyrtus rhodococcusiae Wang & Zhang Encyrtus rhodococcusiae Wang & Zhang in Wang et al. (2016) (nomen nudum) Diagnosis. Encyrtus rhodococcusiae can be diagnosed morphologically within the E. sasakii complex through the combination of the 2" outer plate less than 0.6x as long as the ovipositor shield (2° outer plate at least 0.65x as long as the ovipositor shield in E. eulecaniumiae) and hypopygium deeply concave (hypopygium shallowly concave in E. sasakii). Encyrtus rhodococcusiae can be further differentiated from E. sasakii by having the ventral surface of the clava more than 1.5x as long as the dorsal one (in E. sasakii, the ventral surface of the clava is always less than 1.5x as long as the dorsal one). According to Wang et al. (2016), E. rhodococcusiae can be molecularly distinguished from its closest allies (E. sasakii and E. eulecaniumiae) by having the following nucleotides in the COI marker: 14 (T), 26 (A), 102 (A), 149 (A), 161 (C), 176 (G), 215 (G), 266 (T), 269 (A), 281 (T), 389 (A), 446 (A), 468 (C), 470 (T) 521 (T) and 530 (G). Encyrtus rhodococcusiae is also unique within the E. sasakii complex in attacking the coccid species Rhodococcus sariuoni Borchsenius. Description. (reproduced from Wang et al. 2016) Female — Length including ovi- positor 1.9 mm. Colouration: Head black around ocellar area, frontovertex black; malar space brown; antenna with scape yellowish brown; pedicel and flagellum dark brown; maxillary and labial palpi yellowish brown; pronotum dark brown to black dorsally, laterally pale brown; thorax covered with dark-brown setae; mesoscutum mostly black dorsally, laterally brown; sctuellum black with a broad transverse yellow band and a tuft of black bristles apically; metanotum dark brown; tegula dark brown; mesopleu- ron pale brown; fore and hind coxae brownish yellow (Wang et al. 2016: fig. 7d, f); mid coxa mostly brown (Wang et al. 2016: fig. 7e); legs otherwise dark brown; basal one third of forewing hyaline, infuscate elsewhere; forewing with a series of long bris- tles just below the apical third of submarginal; propodeum brown dorsally, yellowish brown laterally; gaster black; ovipositor sheaths yellow, except apex one-third dark brown (Wang et al. 2016: fig. 7b). Head: frontovertex about half the head width; ocelli forming an obtuse angle (~120°); scrobes quite shallow and M-shaped in frontal view; Integrative taxonomy of the Encyrtus sasakii complex 139 eye at least superficially bare; torulus separated from mouth margin by about one of its own length; toruli separated from each other by about 2.5x their own diameter; antenna with scape subcylindrical, 3.4x as long as broad; pedicel subtriangular, 1.4x as long as broad and as broad as scape. Antenna (Wang et al. 2016: fig. 7a) 13-seg- mented, scape cylindrical; clava 3-segmented, its apex distinctly truncate; mandible plow shaped; clypeus with three to six conspicuous, long, suberect setae; maxillary and labial palpi with 4 and 3 segments respectively. 7horax: Mesoscutum 1.44x wider than long, uniformly convex, setose and finely reticulated, without notauli. Pronotum very short, about one eleventh the mesoscutum length, with polygonal reticulation; scutel- lum about 1.2x as long as broad, sculpture anteriorly similar to that of mesoscutum; forewing (Wang et al. 2016: fig. 7c) about 2.3x as long as broad; costal cell with more than one line of setae dorsally; stigma vein apically curved. Gaster: Hypopygium al- most reaching apex of gaster. Type material. (reproduced from Wang et al. 2016) Holotype — 9, specimen 11- 009A, China, Shandong, Linyi, 9.V.2011, col. Xuemei Yang, ex. Rhodococcus sariuoni on Crataegus pinnatifida (deposited in IZCAS). Paratypes — 22°, China, Beijing, Haidian, 15.V2006, col. Yanzhou Zhang, ex. Rhodococcus sariuoni on Matus spectabilis (deposited in IZCAS); 32. 2, China, Heilongjiang, Harbin, 15.VI.2007, col. Yanzhou Zhang, ex. Rhodococcus sariuoni on Prunus persica (deposited in IZCAS); 622 16, China, Jilin, Changchun, 9.VI.2011, col. Ying Wang, ex. Rhodococcus sariuoni on Pru- nus persica (deposited in IZCAS); 22 9 244, China, Qinghai, Xining, 28.VI.2013, col. Haibin Li, Xubo Wang, Xu Zhang, ex. Rhodococcus sariuoni on Prunus cerasifera (deposited in IZCAS); 72 9 243, China, Shandong, Taian, 9.V.2008, col. Yanzhou Zhang, ex. Rhodococcus sariuoni on Prunus cerasifera (deposited in IZCAS); 29 9, Chi- na, Shaanxi, Xianyang, 9.V.2011, col. Feng Yuan, ex. Rhodococcus sariuoni on Malus sieversii (deposited in IZCAS). Morphometry Antenna: Our ANOVA analyses show that the LW ratios of both surfaces of C1 plus C2 are statistically different (p < 0.001 in both analyses) among E. eulecaniumiae, E. rhodococcusiae and E. sasakii. Linear regressions of the LW ratios of both C1 and C2 also differ (p < 0.05 in both) among species, even though the two ratios are sig- nificantly correlated when phylogenetic constraints are enforced (p < 0.001). Linear regression between the dorsal length and width of the clava also shows significant sta- tistical difference among the three species (p < 0.05). We found further significant dif- ferences between FE. eulecaniumiae and E. rhodococcusiae in the LW ratios of the ventral and dorsal surfaces of Cl (p < 0.05 in both) as well as in the dorsal lengths of both Cl and C2 (p < 0.001 and p < 0.05, respectively). In turn, the width of C2 is statistically different between E. sasakii and E. eulecaniumiae (p < 0.05). Our analyses revealed that the LW ratio of the funicle is statistically different among the three species (p < 0.05) and positively correlated with the total length of antenna (p < 0.001). The LW ratios of Fl, F3 and F5 are each separately correlated 140 Andrey Rudoy et al. / Journal of Hymenoptera Research 90: 129-152 (2022) with the total length of antenna as well (p < 0.001 in all analyses). Also, the lengths and widths of all funicular antennomeres are positively allometric (allometry coefficients 1.7—2.0). Linear regressions, however, show that these relationships are not species- specific (p = 0.05-0.1). Ovipositor: The lengths of all analysed components of the ovipositor are signifi- cantly different between pairs of species, even though none of them separates all three species (see Suppl. material 1: Table $3). On the other hand, the length ratio between the 1“ and 2™ outer plates (p < 0.001) as well as that between the lengths of the 2"* out- er plate and shield of the ovipositor (p < 0.001) are both significantly different among the three species. Our analyses also show that (i) the length ratio between the two outer plates is significantly correlated with the length of the ovipositor shield (p < 0.001); (ii) the relationship between length and width is positively allometric in both the sty- lus and 2" outer plate of the ovipositor, slopes 1.67 (p < 0.001) and 2.5 (p < 0.001), respectively; and (iii) the lengths of the stylus and ovipositor shield are also allometric (slope 2.0, p < 0.001). None of these results, however, shows species separation. Among the analysed components of the ovipositor, the shape (EPC1) of the shield differs significantly among the three species (p < 0.001). The shapes of the two out- er plates, however, differ only between E. rhodococcusiae and the other two species (p < 0.01), although linear regression of the shape parameters of the 2"¢ outer plate pro- vides statistical difference among all species (p < 0.05). According to EPC1, the shapes of the two outer plates of the ovipositor are correlated with each other (p < 0.005), as is the shape of the ovipositor shield and its total length (p < 0.001), in all three species. Hypopygium: PC2 of hypopygium shape shows clear separation among all three species (p < 0.001). In all of them, the shape of the hypopygium (PC1) is allometric in relation to its barycenter size (ordinary correlation: p < 0.001, evolutionary correla- tion: p < 0.05). Regression between hypopygium shape (PC1) and length of ovipositor shield shows difference among the species (p < 0.05). Both EPC1 and EPC2 indi- cate that the shape parameters of the hypopygium are correlated with forewing length (p < 0.05 and p < 0.001, respectively). Correlations between structures: The lengths of the antenna, hypopygium and all the components of the ovipositor are negatively allometric to forewing length (p < 0.001). In turn, the forewing length is positively correlated with the LW ratio of clava (p < 0.001) as well as with EPC1 of the shapes of ovipositor shield and 2™ outer plate (p < 0.001 and p < 0.05, respectively). The total length of antenna and ventral length of C2 are independently positively correlated with EPC2 of the hypopygium shape (p < 0.001 and p < 0.05, respectively). The average LW ratio of the ventral sur- face of Cl plus C2 is strongly correlated with the shape (EPC1) of the 2°¢ outer plate of the ovipositor (p < 0.001). EPC1 of the shapes of the hypopygium and ovipositor shield are also correlated to each other (p < 0.05). Phylogenetic analysis: The majority-rule consensus trees obtained through the Bayesian analysis of the combined dataset, as well as those of the 28S and COI datasets only, are shown in Figs 3, $1, S2, respectively. The two latter trees show that each of the three species belonging to the £. sasakii complex forms a monophyletic group, Integrative taxonomy of the Encyrtus sasakii complex 98 7 100/57 ~N 99 100Lfgg 100 98 70 100 T00 S61 00 E. sasakii complex 100 100 871 I 75 99 (51 82 “J 2) (oe) os co) oO mmm mn mm mmm mm mm mm mmm mmm mim mm 821100 (